|View source on GitHub|
Represents a model for use in TensorFlow Federated.
Model will work on a set of
tf.Variables, and each method should be
a computation that can be implemented as a
tf.function; this implies the
class should essentially be stateless from a Python perspective, as each
method will generally only be traced once (per set of arguments) to create the
corresponding TensorFlow graph functions. Thus,
Model instances should
behave as expected in both eager and graph (TF 1.0) usage.
tf.Variables may be either:
- Weights, the variables needed to make predictions with the model.
- Local variables, e.g. to accumulate aggregated metrics across calls to forward_pass.
The weights can be broken down into trainable variables (variables
that can and should be trained using gradient-based methods), and
non-trainable variables (which could include fixed pre-trained layers,
or static model data). These variables are provided via the
properties, and must be initialized by the user of the
In federated learning, model weights will generally be provided by the
server, and updates to trainable model variables will be sent back to the
server. Local variables are not transmitted, and are instead initialized
locally on the device, and then used to produce
are sent to the server.
tf.Variables should be introduced in
__init__; this could move to a
build method more inline with Keras (see
Performs federated aggregation of the
This is typically used to aggregate metrics across many clients, e.g. the body of the computation might be:
N.B. It is assumed all TensorFlow computation happens in the
The type specification of the
A nested structure of
Similar in spirit to
An iterable of
An iterable of
An iterable of
forward_pass( batch_input, training=True )
Runs the forward pass and returns results.
This method should not modify any variables that are part of the model parameters, that is, variables that influence the predictions. Rather, this is done by the training loop.
However, this method may update aggregated metrics computed across calls to
forward_pass; the final values of such metrics can be accessed via
Uses in TFF:
- To implement model evaluation.
- To implement federated gradient descent and other non-Federated-Averaging algorithms, where we want the model to run the forward pass and update metrics, but there is no optimizer (we might only compute gradients on the returned loss).
- To implement Federated Averaging.
a nested structure that matches the structure of
Returns tensors representing values aggregated over
In federated learning, the values returned by this method will typically be further aggregated across clients and made available on the server.
This method returns results from aggregating across all previous calls
forward_pass, most typically metrics like accuracy and loss. If needed,
we may add a
clear_aggregated_outputs method, which would likely just
run the initializers on the
In general, the tensors returned can be an arbitrary function of all
tf.Variables of this model, not just the
example, this could return tensors measuring the total L2 norm of the model
(which might have been updated by training).
This method may return arbitrarily shaped tensors, not just scalar metrics. For example, it could return the average feature vector or a count of how many times each feature exceed a certain magnitude.
A structure of tensors (as supported by